Wednesday, September 30, 2020

How accurate are the different coronavirus tests?

I looked recently into the accuracy of forehead thermometers and, while I concluded that they were useful as an initial screening tool, they are not equivalent to an actual COVID-19 test.

There are two main types of tests for the coronavirus: the molecular real-time polymerase chain reaction nasopharygeal  test (usually shortened to "RT-PCR test" or just "PCR test", for obvious reaons), and the antigen test. Antibody (or serology) tests also exist, but they do not indicate an active virus, and are of more limited use for diagnosis. Some 130 different tests have been approved by the US's Federal Drug Agency (FDA) under its Emergency Use Authorization (EUA) protocol, which allows '"unapproved uses of approved medical products" in emergency situations like this, but none of them are fully approved by the FDA, with all the rigorous testing and vetting of new procedures that normally happens. And, as we shall see, some tests are definitely better than others.

The PCR test is still considered the gold standard, and here we are talking about the familiar image from countless television reports of a white-coated healthcare worker poking a long swab deep into a testee's nose. The test searches for the presence of the virus' genetic material deep in a person's nasal cavity. Done properly (i.e. deep enough to cause discomfort, to where the nasal cavity meet the pharynx at the back of the mouth), the PCR test is almost 100% accurate and effective, even in the absence of symptoms. 

However, ideally, the test should be done least 8 days after infection to ensure that sufficient viral particles are present (i.e. a test performed soon after infection is less effective and more likely to produce a false negative result). Less invasive, more shallow swabs of the nose, or of saliva from the back of the mouth,which have been authorized recently in the US, are also typically less accurate, although still quite good. False positive results, where viral particles remain long after a person has recovered from the disease, are possible, but not that common.

PCR tests must be analyzed in a lab on large specialized equipment, and it can take several days for results to become available, or even up to a week or more (theoretically, they should be available the same day, but in practice this does not happen). The other issue that is arising with extensive PCR testing is a shortage of reagents, the chemicals that are needed to analyze the tests, which in many cases is proving to be the limiting factor on how many and how fast tests can be performed. "Pooled testing" may speed things up, whereby batches of tests are analyzed together: if the result is negative, no more needs to be done, while, if the pool result is positive, individual tests can then be re-analyzed. 

The FDA has also recently approved saliva or spit tests under its EUA protocol, such as the SalivaDirect test developed by the Yale School of Public Health, among others. This is a PCR test, but using the quick and simple expedient of spitting into a container, rather than the invasive and time-consuming nasopharyngeal swab method using trained personnel. These newer tests appear to be very accurate (Yale claims 94% accuracy compared to nasopharyngeal PCR tests), as well as quicker, cheaper and less instrusive, and home kits are expected to be available soon.

The other main COVID diagnostic test, antigen testing, on the other hand, produces results within just a few hours, even minutes. However, the compromise is that that they are substantially less accurate, mainly because they require large amounts of viral material to yield a positive result. Antigen tests use nasal or throat swabs to look for the presence of certain proteins on the surface of the virus. If the test is positive, it is generally reliable, but false negatives, where the virus is present but not picked up, can occur in 20-30% of tests, even as high as 50%, making it much less effective than the PCR test.

There are also antibody tests available. Antibody tests are not diagnostic tests: they do not indicate that someone has the virus, merely that they had it at some point. They require blood samples not nasal swabs, and results are typically available within a few days, or less. 

However, COVID-19 antibodies can take several days or even weeks to develop after an infection, and the jury is still out on how long they remain active (probably just several weeks), and to what extent they would be effective on a mutating virus. Long-term T-cell analysis would be better in this regard, but no such test is easily or commercially available. 

Antigen tests are subject to false positive results, as the antibodies may occur in response to a different coronavirus. False negative rates of 20-30% have also been reported. So, the antibody test is not considered very useful or reliable, and may give a false sense of security. The presence of antibodies may not mean that a person is immune to a different (or even the same) strain of the virus.

So, it's PCR all the way, despite the slowness of results and the existence of other alternative tests, which are better used as confirmatory tests. But new saliva-based PCR tests are much quicker and cheaper, less invasive, and almost as reliable, and may be the way forward. Amid reports of 6-7 hour-long waits for traditional testing, this may be just what we need, but don't expect Canada's highly cautious and creaky old approval system to authorize them any time soon.

UPDATE

Well, I shouldn't speak too soon. Canada has just announcing that it is ordering nearly 8 million rapid tests from Abbott Rapid Diagnostics ULC, despite Health Canada not having approved the technology yet.

ID Now is a PCR test that has been available in the USA since March, but it yields results in just 5-13 minutes using a portable machine the size of a toaster. It still requires a qualified  health professional to collect a sample using a throat or nasal swab, but it could nevertheless speed up testing significantly. There was some concern early on in the US that the tests yielded many false negative results, but Abbott maintains that these problems have been fixed and that tests perform well when administered soon after symptoms appear.

UPDATE UPDATE

Health Canada has finally got around to granting its official approval to Abbott ID Now test

Now, all we need are the actualy test kits...

No comments: